Revert "Fix rotation around Y axis"

[johnmccutchan]

Reverts #262

• Broke some tests
• Made the method inconsistent with the Matrix4 code.

[tlserver]

Hi @johnmccutchan,

After reviewing the revert on PR #262 regarding the "Fix rotation around Y axis" and considering the rotation matrix conventions as explained in the Wikipedia article on rotation matrices (https://en.wikipedia.org/wiki/Rotation_matrix#Nested_dimensions), I have some concerns about the revert.

The documentation suggests that for a rotation around the Y-axis in a right-handed coordinate system, one of the sine values should indeed be negated. This is to ensure that the transformation adheres to the standard definition of a rotation matrix in Euclidean space.

I understand that the revert was prompted by failing tests and by making the method inconsistent with the Matrix4 code. However, if the tests are failing due to the correction aligning with mathematical standards, then it could indicate that the tests may need to be revised. Moreover, the inconsistency with the Matrix4 code should also be scrutinized, because our goal should be to ensure that all parts of the codebase accurately reflect the correct mathematical operations.

It's crucial for the integrity of the library that all rotation-related methods and representations remain consistent with one another and with standard mathematical definitions. To this end, I recommend a comprehensive review of both the test cases and the Matrix4 implementation. If the Matrix4 code is indeed correct, then the method for rotation around the Y-axis should be consistent with that implementation as long as both adhere to the correct mathematical principles.

Perhaps we could collaborate on identifying the discrepancies and updating the affected tests and code? Ensuring consistency across our methods while upholding the standards of rotation matrix mathematics will greatly enhance the reliability and predictability of the library.

Looking forward to your thoughts on this matter.